Cursor's New Plans Include Grok. Claude and GPT Now Bill at API List Price

• Book: AI Agents Pocket Guide
• Also by me: Thinking in Go (2-book series) — Complete Guide to Go Programming + Hexagonal Architecture in Go
• My project: Hermes IDE | GitHub — an IDE for developers who ship with Claude Code and other AI coding tools
• Me: xgabriel.com | GitHub

If you sign up for Cursor this week, the plan you buy doesn't include Claude or GPT. Since October 12, new Cursor customers start on The X Subscription, a bundle of Grok, Cursor and X Premium whose included usage covers Grok models.

Cursor's help page puts it plainly: "third-party models in Cursor, like Claude or GPT, aren't included in your plan." On Premium and above you can still pick them. You pay API list prices from the first request, out of prepaid credit.

Existing subscribers are not moved, so which group you're in decides whether you need to do anything today.

What changed on October 12

The bundle comes from SpaceXSI, which TechCrunch calls "the company now known as SpaceX Super Intelligence." It announced The X Subscription on x.ai on October 12, and Cursor posted it to its community forum the same day. SpaceX agreed in June to buy Anysphere, the company behind Cursor, for $60 billion, Techzine reported.

The individual tiers, from Cursor's pricing page:

Teams start at $30 per user per month, with 4x to 50x Plus usage per seat (pricing).

Paid plans share one weekly pool between Grok and Cursor (help page). A Grok chat and a Cursor agent run draw from the same allowance. The pricing page states usage only as multiples of Plus, not as tokens or requests, so you can't yet convert "4x" into a number of agent sessions.

What "API list price from first use" means for your bill

The pricing page FAQ says Premium and above "do not include an allowance for third-party models" and that you "pay for exactly what you use," with optional spending caps (pricing).

So the plan price and your Claude bill become two separate numbers. Anthropic's list prices are $2 per million input tokens and $10 per million output for Sonnet 5.5, and $4 / $20 for Opus 5.5. Here is the arithmetic for an illustrative session I made up:

type Price = { input: number; output: number };

// USD per 1M tokens, Anthropic list prices
const SONNET_5_5: Price = { input: 2, output: 10 };
const OPUS_5_5: Price = { input: 4, output: 20 };

function listCost(p: Price, inTok: number, outTok: number) {
  return (inTok * p.input + outTok * p.output) / 1_000_000;
}

listCost(SONNET_5_5, 3_000_000, 150_000); // 7.5
listCost(OPUS_5_5, 3_000_000, 150_000); // 15

Three million uncached input tokens and 150,000 output tokens come to $7.50 on Sonnet 5.5 and $15 on Opus 5.5. A coding agent resends its conversation on every turn, which is why input tokens pile up fast. Anthropic lists cache reads far lower, at $0.10 per million on Sonnet 5.5 (pricing), so cache hits can move the total a lot. Cursor's pages say "API list prices" and don't spell out how caching is billed, so treat the function as an estimate, not a quote.

Set the spending cap before your first Claude request, while your prepaid credit is still untouched.

Inside the plan: one weekly pool for Grok and Cursor. Outside it: third-party models, metered from the first request.

Who this hits, and who it doesn't

• Existing subscribers: your current Cursor, SuperGrok or X Premium plan, price and features continue unless you change or cancel (help page). The forum announcement says current Cursor subscribers keep their included third-party usage until they switch.
• New customers: they start on The X Subscription, "including anyone with only a free X account" (help page). In the forum thread, the announcement's author says the new Ultra is a different plan from the previous Cursor Ultra, which is no longer open to new signups.
• Existing plan changes: you can still move between your current Cursor plans without switching, except plans billed through the Apple App Store, which have to switch first (help page).
• Teams and Enterprise: existing teams can add seats or change a seat's tier without switching. Enterprise customers without an active Grok account can switch with help from their account team; other Enterprise accounts can't switch yet (xAI docs).

Then there's the line to read twice before you click anything in billing settings: "The switch is permanent. It can't be undone, and Support can't undo it for you" (help page). If you switch and your Cursor plan came from the App Store, you have to cancel it in your device settings yourself, or Apple keeps billing you.

Switching comes with a prorated refund for your old plan and limited-time offers shown in your billing settings (x.ai). If most of your Cursor work runs on Claude or GPT today, price last month's usage at list rates before you take either.

The case for the bundle

The deal is real if Grok models do your work well. SpaceXSI says Premium bundles SuperGrok, Cursor Pro and X Premium+, which it puts at $90 a month combined, for $30 (x.ai). It also says "Each X Subscription plan includes at least twice as much coding usage with our models as the comparable Cursor plan," and that each comparable tier "provides more usage-per-dollar than the plan it replaces." Premium lists cloud agents and automations, code hosting and AI code reviews (pricing).

The early reaction is harsher. Several replies in the forum thread are critical. One reads: "Cursor's users came for the freedom to choose their model." A forum thread is a small, self-selected sample, so read that as a mood.

I haven't run Grok inside Cursor myself, and I can't tell you how it does on your codebase. That's the test worth running before you pick a tier: same ticket, Grok in Cursor, and the model you use today, then compare the diffs and the bill.

The trade: more bundled usage per dollar, if the included models are the ones you want.

Keep the model choice in your hands

This is the second time on October 12 that a plan decided which model you can reach. The same day, Meta said its larger Muse Wonder Full model will be available only to Muse Code subscribers and won't be served through its API (TechCrunch). Now Cursor's included usage is Grok's.

Three things to do this week:

1. List the models your team uses in Cursor and what that usage would cost at list price.
2. Turn on a spending cap for third-party usage if you're on a new plan.
3. Keep at least one agent whose own plan includes the model you depend on, so a pricing change somewhere else doesn't stop your work.

I build Hermes IDE, so weigh this with that in mind. Hermes runs Claude Code, Codex, Copilot, OpenCode and other command-line agents side by side, each in its own terminal and its own worktree, with your own sign-ins (GitHub). You pick the agent, model, effort and account per task, and when one agent hits its usage limit, you can continue the task in another. It won't make anyone's tokens cheaper. What it changes is how much a pricing change at one vendor costs you: you move the session to another agent and keep going. It's free on macOS, Windows and Linux (download), source-available under BSL 1.1, and not affiliated with Nous Research, Anthropic, OpenAI or SpaceXSI.

So, would you switch?

If you use Cursor every day: does $30 of bundled Grok usage beat paying list price for the Claude or GPT tokens you already burn? Or is a plan that only covers one vendor's models a dealbreaker? Tell me in the comments.

If this was useful

I build Hermes IDE so you can run the agents you already pay for side by side, each on its own branch, and move a task to another agent when a plan runs out. If that matters to you, star it on GitHub. If you build agents yourself, the AI Agents Pocket Guide covers the patterns behind tools like these.